lib/Support/PathV2.cpp

   1 //===-- PathV2.cpp - Implement OS Path Concept ------------------*- C++ -*-===//
   2 //
   3 //                     The LLVM Compiler Infrastructure
   4 //
   5 // This file is distributed under the University of Illinois Open Source
   6 // License. See LICENSE.TXT for details.
   7 //
   8 //===----------------------------------------------------------------------===//
   9 //
  10 //  This file implements the operating system PathV2 API.
  11 //
  12 //===----------------------------------------------------------------------===//
  13
  14 #include "llvm/Support/PathV2.h"
  15 #include "llvm/Support/FileSystem.h"
  16 #include "llvm/Support/ErrorHandling.h"
  17 #include <cctype>
  18 #include <cstdio>
  19 #include <cstring>
  20
  21 namespace {
  22   using llvm::StringRef;
  23   using llvm::sys::path::is_separator;
  24
  25 #ifdef LLVM_ON_WIN32
  26   const StringRef separators = "\\/";
  27   const char      prefered_separator = '\\';
  28 #else
  29   const StringRef separators = "/";
  30   const char      prefered_separator = '/';
  31 #endif
  32
  33   const llvm::error_code success;
  34
  35   StringRef find_first_component(StringRef path) {
  36     // Look for this first component in the following order.
  37     // * empty (in this case we return an empty string)
  38     // * either C: or {//,\\}net.
  39     // * {/,\}
  40     // * {.,..}
  41     // * {file,directory}name
  42
  43     if (path.empty())
  44       return path;
  45
  46 #ifdef LLVM_ON_WIN32
  47     // C:
  48     if (path.size() >= 2 && std::isalpha(path[0]) && path[1] == ':')
  49       return path.substr(0, 2);
  50 #endif
  51
  52     // //net
  53     if ((path.size() > 2) &&
  54         is_separator(path[0]) &&
  55         path[0] == path[1] &&
  56         !is_separator(path[2])) {
  57       // Find the next directory separator.
  58       size_t end = path.find_first_of(separators, 2);
  59       return path.substr(0, end);
  60     }
  61
  62     // {/,\}
  63     if (is_separator(path[0]))
  64       return path.substr(0, 1);
  65
  66     if (path.startswith(".."))
  67       return path.substr(0, 2);
  68
  69     if (path[0] == '.')
  70       return path.substr(0, 1);
  71
  72     // * {file,directory}name
  73     size_t end = path.find_first_of(separators, 2);
  74     return path.substr(0, end);
  75   }
  76
  77   size_t filename_pos(StringRef str) {
  78     if (str.size() == 2 &&
  79         is_separator(str[0]) &&
  80         str[0] == str[1])
  81       return 0;
  82
  83     if (str.size() > 0 && is_separator(str[str.size() - 1]))
  84       return str.size() - 1;
  85
  86     size_t pos = str.find_last_of(separators, str.size() - 1);
  87
  88 #ifdef LLVM_ON_WIN32
  89     if (pos == StringRef::npos)
  90       pos = str.find_last_of(':', str.size() - 2);
  91 #endif
  92
  93     if (pos == StringRef::npos ||
  94         (pos == 1 && is_separator(str[0])))
  95       return 0;
  96
  97     return pos + 1;
  98   }
  99
 100   size_t root_dir_start(StringRef str) {
 101     // case "c:/"
 102 #ifdef LLVM_ON_WIN32
 103     if (str.size() > 2 &&
 104         str[1] == ':' &&
 105         is_separator(str[2]))
 106       return 2;
 107 #endif
 108
 109     // case "//"
 110     if (str.size() == 2 &&
 111         is_separator(str[0]) &&
 112         str[0] == str[1])
 113       return StringRef::npos;
 114
 115     // case "//net"
 116     if (str.size() > 3 &&
 117         is_separator(str[0]) &&
 118         str[0] == str[1] &&
 119         !is_separator(str[2])) {
 120       return str.find_first_of(separators, 2);
 121     }
 122
 123     // case "/"
 124     if (str.size() > 0 && is_separator(str[0]))
 125       return 0;
 126
 127     return StringRef::npos;
 128   }
 129
 130   size_t parent_path_end(StringRef path) {
 131     size_t end_pos = filename_pos(path);
 132
 133     bool filename_was_sep = path.size() > 0 && is_separator(path[end_pos]);
 134
 135     // Skip separators except for root dir.
 136     size_t root_dir_pos = root_dir_start(path.substr(0, end_pos));
 137
 138     while(end_pos > 0 &&
 139           (end_pos - 1) != root_dir_pos &&
 140           is_separator(path[end_pos - 1]))
 141       --end_pos;
 142
 143     if (end_pos == 1 && root_dir_pos == 0 && filename_was_sep)
 144       return StringRef::npos;
 145
 146     return end_pos;
 147   }
 148 } // end unnamed namespace
 149
 150 namespace llvm {
 151 namespace sys  {
 152 namespace path {
 153
 154 const_iterator begin(StringRef path) {
 155   const_iterator i;
 156   i.Path      = path;
 157   i.Component = find_first_component(path);
 158   i.Position  = 0;
 159   return i;
 160 }
 161
 162 const_iterator end(StringRef path) {
 163   const_iterator i;
 164   i.Path      = path;
 165   i.Position  = path.size();
 166   return i;
 167 }
 168
 169 const_iterator &const_iterator::operator++() {
 170   assert(Position < Path.size() && "Tried to increment past end!");
 171
 172   // Increment Position to past the current component
 173   Position += Component.size();
 174
 175   // Check for end.
 176   if (Position == Path.size()) {
 177     Component = StringRef();
 178     return *this;
 179   }
 180
 181   // Both POSIX and Windows treat paths that begin with exactly two separators
 182   // specially.
 183   bool was_net = Component.size() > 2 &&
 184     is_separator(Component[0]) &&
 185     Component[1] == Component[0] &&
 186     !is_separator(Component[2]);
 187
 188   // Handle separators.
 189   if (is_separator(Path[Position])) {
 190     // Root dir.
 191     if (was_net
 192 #ifdef LLVM_ON_WIN32
 193         // c:/
 194         || Component.endswith(":")
 195 #endif
 196         ) {
 197       Component = Path.substr(Position, 1);
 198       return *this;
 199     }
 200
 201     // Skip extra separators.
 202     while (Position != Path.size() &&
 203            is_separator(Path[Position])) {
 204       ++Position;
 205     }
 206
 207     // Treat trailing '/' as a '.'.
 208     if (Position == Path.size()) {
 209       --Position;
 210       Component = ".";
 211       return *this;
 212     }
 213   }
 214
 215   // Find next component.
 216   size_t end_pos = Path.find_first_of(separators, Position);
 217   Component = Path.slice(Position, end_pos);
 218
 219   return *this;
 220 }
 221
 222 const_iterator &const_iterator::operator--() {
 223   // If we're at the end and the previous char was a '/', return '.'.
 224   if (Position == Path.size() &&
 225       Path.size() > 1 &&
 226       is_separator(Path[Position - 1])
 227 #ifdef LLVM_ON_WIN32
 228       && Path[Position - 2] != ':'
 229 #endif
 230       ) {
 231     --Position;
 232     Component = ".";
 233     return *this;
 234   }
 235
 236   // Skip separators unless it's the root directory.
 237   size_t root_dir_pos = root_dir_start(Path);
 238   size_t end_pos = Position;
 239
 240   while(end_pos > 0 &&
 241         (end_pos - 1) != root_dir_pos &&
 242         is_separator(Path[end_pos - 1]))
 243     --end_pos;
 244
 245   // Find next separator.
 246   size_t start_pos = filename_pos(Path.substr(0, end_pos));
 247   Component = Path.slice(start_pos, end_pos);
 248   Position = start_pos;
 249   return *this;
 250 }
 251
 252 bool const_iterator::operator==(const const_iterator &RHS) const {
 253   return Path.begin() == RHS.Path.begin() &&
 254          Position == RHS.Position;
 255 }
 256
 257 bool const_iterator::operator!=(const const_iterator &RHS) const {
 258   return !(*this == RHS);
 259 }
 260
 261 ptrdiff_t const_iterator::operator-(const const_iterator &RHS) const {
 262   return Position - RHS.Position;
 263 }
 264
 265 const StringRef root_path(StringRef path) {
 266   const_iterator b = begin(path),
 267                  pos = b,
 268                  e = end(path);
 269   if (b != e) {
 270     bool has_net = b->size() > 2 && is_separator((*b)[0]) && (*b)[1] == (*b)[0];
 271     bool has_drive =
 272 #ifdef LLVM_ON_WIN32
 273       b->endswith(":");
 274 #else
 275       false;
 276 #endif
 277
 278     if (has_net || has_drive) {
 279       if ((++pos != e) && is_separator((*pos)[0])) {
 280         // {C:/,//net/}, so get the first two components.
 281         return path.substr(0, b->size() + pos->size());
 282       } else {
 283         // just {C:,//net}, return the first component.
 284         return *b;
 285       }
 286     }
 287
 288     // POSIX style root directory.
 289     if (is_separator((*b)[0])) {
 290       return *b;
 291     }
 292   }
 293
 294   return StringRef();
 295 }
 296
 297 const StringRef root_name(StringRef path) {
 298   const_iterator b = begin(path),
 299                  e = end(path);
 300   if (b != e) {
 301     bool has_net = b->size() > 2 && is_separator((*b)[0]) && (*b)[1] == (*b)[0];
 302     bool has_drive =
 303 #ifdef LLVM_ON_WIN32
 304       b->endswith(":");
 305 #else
 306       false;
 307 #endif
 308
 309     if (has_net || has_drive) {
 310       // just {C:,//net}, return the first component.
 311       return *b;
 312     }
 313   }
 314
 315   // No path or no name.
 316   return StringRef();
 317 }
 318
 319 const StringRef root_directory(StringRef path) {
 320   const_iterator b = begin(path),
 321                  pos = b,
 322                  e = end(path);
 323   if (b != e) {
 324     bool has_net = b->size() > 2 && is_separator((*b)[0]) && (*b)[1] == (*b)[0];
 325     bool has_drive =
 326 #ifdef LLVM_ON_WIN32
 327       b->endswith(":");
 328 #else
 329       false;
 330 #endif
 331
 332     if ((has_net || has_drive) &&
 333         // {C:,//net}, skip to the next component.
 334         (++pos != e) && is_separator((*pos)[0])) {
 335       return *pos;
 336     }
 337
 338     // POSIX style root directory.
 339     if (!has_net && is_separator((*b)[0])) {
 340       return *b;
 341     }
 342   }
 343
 344   // No path or no root.
 345   return StringRef();
 346 }
 347
 348 const StringRef relative_path(StringRef path) {
 349   StringRef root = root_path(path);
 350   return root.substr(root.size());
 351 }
 352
 353 void append(SmallVectorImpl<char> &path, const Twine &a,
 354                                          const Twine &b,
 355                                          const Twine &c,
 356                                          const Twine &d) {
 357   SmallString<32> a_storage;
 358   SmallString<32> b_storage;
 359   SmallString<32> c_storage;
 360   SmallString<32> d_storage;
 361
 362   SmallVector<StringRef, 4> components;
 363   if (!a.isTriviallyEmpty()) components.push_back(a.toStringRef(a_storage));
 364   if (!b.isTriviallyEmpty()) components.push_back(b.toStringRef(b_storage));
 365   if (!c.isTriviallyEmpty()) components.push_back(c.toStringRef(c_storage));
 366   if (!d.isTriviallyEmpty()) components.push_back(d.toStringRef(d_storage));
 367
 368   for (SmallVectorImpl<StringRef>::const_iterator i = components.begin(),
 369                                                   e = components.end();
 370                                                   i != e; ++i) {
 371     bool path_has_sep = !path.empty() && is_separator(path[path.size() - 1]);
 372     bool component_has_sep = !i->empty() && is_separator((*i)[0]);
 373     bool is_root_name = has_root_name(*i);
 374
 375     if (path_has_sep) {
 376       // Strip separators from beginning of component.
 377       size_t loc = i->find_first_not_of(separators);
 378       StringRef c = i->substr(loc);
 379
 380       // Append it.
 381       path.append(c.begin(), c.end());
 382       continue;
 383     }
 384
 385     if (!component_has_sep && !(path.empty() || is_root_name)) {
 386       // Add a separator.
 387       path.push_back(prefered_separator);
 388     }
 389
 390     path.append(i->begin(), i->end());
 391   }
 392 }
 393
 394 void append(SmallVectorImpl<char> &path,
 395             const_iterator begin, const_iterator end) {
 396   for (; begin != end; ++begin)
 397     path::append(path, *begin);
 398 }
 399
 400 const StringRef parent_path(StringRef path) {
 401   size_t end_pos = parent_path_end(path);
 402   if (end_pos == StringRef::npos)
 403     return StringRef();
 404   else
 405     return path.substr(0, end_pos);
 406 }
 407
 408 void remove_filename(SmallVectorImpl<char> &path) {
 409   size_t end_pos = parent_path_end(StringRef(path.begin(), path.size()));
 410   if (end_pos != StringRef::npos)
 411     path.set_size(end_pos);
 412 }
 413
 414 void replace_extension(SmallVectorImpl<char> &path, const Twine &extension) {
 415   StringRef p(path.begin(), path.size());
 416   SmallString<32> ext_storage;
 417   StringRef ext = extension.toStringRef(ext_storage);
 418
 419   // Erase existing extension.
 420   size_t pos = p.find_last_of('.');
 421   if (pos != StringRef::npos && pos >= filename_pos(p))
 422     path.set_size(pos);
 423
 424   // Append '.' if needed.
 425   if (ext.size() > 0 && ext[0] != '.')
 426     path.push_back('.');
 427
 428   // Append extension.
 429   path.append(ext.begin(), ext.end());
 430 }
 431
 432 void native(const Twine &path, SmallVectorImpl<char> &result) {
 433   // Clear result.
 434   result.clear();
 435 #ifdef LLVM_ON_WIN32
 436   SmallString<128> path_storage;
 437   StringRef p = path.toStringRef(path_storage);
 438   result.reserve(p.size());
 439   for (StringRef::const_iterator i = p.begin(),
 440                                  e = p.end();
 441                                  i != e;
 442                                  ++i) {
 443     if (*i == '/')
 444       result.push_back('\\');
 445     else
 446       result.push_back(*i);
 447   }
 448 #else
 449   path.toVector(result);
 450 #endif
 451 }
 452
 453 const StringRef filename(StringRef path) {
 454   return *(--end(path));
 455 }
 456
 457 const StringRef stem(StringRef path) {
 458   StringRef fname = filename(path);
 459   size_t pos = fname.find_last_of('.');
 460   if (pos == StringRef::npos)
 461     return fname;
 462   else
 463     if ((fname.size() == 1 && fname == ".") ||
 464         (fname.size() == 2 && fname == ".."))
 465       return fname;
 466     else
 467       return fname.substr(0, pos);
 468 }
 469
 470 const StringRef extension(StringRef path) {
 471   StringRef fname = filename(path);
 472   size_t pos = fname.find_last_of('.');
 473   if (pos == StringRef::npos)
 474     return StringRef();
 475   else
 476     if ((fname.size() == 1 && fname == ".") ||
 477         (fname.size() == 2 && fname == ".."))
 478       return StringRef();
 479     else
 480       return fname.substr(pos);
 481 }
 482
 483 bool is_separator(char value) {
 484   switch(value) {
 485 #ifdef LLVM_ON_WIN32
 486     case '\\': // fall through
 487 #endif
 488     case '/': return true;
 489     default: return false;
 490   }
 491 }
 492
 493 bool has_root_name(const Twine &path) {
 494   SmallString<128> path_storage;
 495   StringRef p = path.toStringRef(path_storage);
 496
 497   return !root_name(p).empty();
 498 }
 499
 500 bool has_root_directory(const Twine &path) {
 501   SmallString<128> path_storage;
 502   StringRef p = path.toStringRef(path_storage);
 503
 504   return !root_directory(p).empty();
 505 }
 506
 507 bool has_root_path(const Twine &path) {
 508   SmallString<128> path_storage;
 509   StringRef p = path.toStringRef(path_storage);
 510
 511   return !root_path(p).empty();
 512 }
 513
 514 bool has_relative_path(const Twine &path) {
 515   SmallString<128> path_storage;
 516   StringRef p = path.toStringRef(path_storage);
 517
 518   return !relative_path(p).empty();
 519 }
 520
 521 bool has_filename(const Twine &path) {
 522   SmallString<128> path_storage;
 523   StringRef p = path.toStringRef(path_storage);
 524
 525   return !filename(p).empty();
 526 }
 527
 528 bool has_parent_path(const Twine &path) {
 529   SmallString<128> path_storage;
 530   StringRef p = path.toStringRef(path_storage);
 531
 532   return !parent_path(p).empty();
 533 }
 534
 535 bool has_stem(const Twine &path) {
 536   SmallString<128> path_storage;
 537   StringRef p = path.toStringRef(path_storage);
 538
 539   return !stem(p).empty();
 540 }
 541
 542 bool has_extension(const Twine &path) {
 543   SmallString<128> path_storage;
 544   StringRef p = path.toStringRef(path_storage);
 545
 546   return !extension(p).empty();
 547 }
 548
 549 bool is_absolute(const Twine &path) {
 550   SmallString<128> path_storage;
 551   StringRef p = path.toStringRef(path_storage);
 552
 553   bool rootDir = has_root_directory(p),
 554 #ifdef LLVM_ON_WIN32
 555        rootName = has_root_name(p);
 556 #else
 557        rootName = true;
 558 #endif
 559
 560   return rootDir && rootName;
 561 }
 562
 563 bool is_relative(const Twine &path) {
 564   return !is_absolute(path);
 565 }
 566
 567 } // end namespace path
 568
 569 namespace fs {
 570
 571 error_code make_absolute(SmallVectorImpl<char> &path) {
 572   StringRef p(path.data(), path.size());
 573
 574   bool rootName      = path::has_root_name(p),
 575        rootDirectory = path::has_root_directory(p);
 576
 577   // Already absolute.
 578   if (rootName && rootDirectory)
 579     return success;
 580
 581   // All of the following conditions will need the current directory.
 582   SmallString<128> current_dir;
 583   if (error_code ec = current_path(current_dir)) return ec;
 584
 585   // Relative path. Prepend the current directory.
 586   if (!rootName && !rootDirectory) {
 587     // Append path to the current directory.
 588     path::append(current_dir, p);
 589     // Set path to the result.
 590     path.swap(current_dir);
 591     return success;
 592   }
 593
 594   if (!rootName && rootDirectory) {
 595     StringRef cdrn = path::root_name(current_dir);
 596     SmallString<128> curDirRootName(cdrn.begin(), cdrn.end());
 597     path::append(curDirRootName, p);
 598     // Set path to the result.
 599     path.swap(curDirRootName);
 600     return success;
 601   }
 602
 603   if (rootName && !rootDirectory) {
 604     StringRef pRootName      = path::root_name(p);
 605     StringRef bRootDirectory = path::root_directory(current_dir);
 606     StringRef bRelativePath  = path::relative_path(current_dir);
 607     StringRef pRelativePath  = path::relative_path(p);
 608
 609     SmallString<128> res;
 610     path::append(res, pRootName, bRootDirectory, bRelativePath, pRelativePath);
 611     path.swap(res);
 612     return success;
 613   }
 614
 615   llvm_unreachable("All rootName and rootDirectory combinations should have "
 616                    "occurred above!");
 617 }
 618
 619 error_code create_directories(const Twine &path, bool &existed) {
 620   SmallString<128> path_storage;
 621   StringRef p = path.toStringRef(path_storage);
 622
 623   StringRef parent = path::parent_path(p);
 624   bool parent_exists;
 625
 626   if (error_code ec = fs::exists(parent, parent_exists)) return ec;
 627
 628   if (!parent_exists)
 629     return create_directories(parent, existed);
 630
 631   return create_directory(p, existed);
 632 }
 633
 634 bool exists(file_status status) {
 635   return status_known(status) && status.type() != file_type::file_not_found;
 636 }
 637
 638 bool status_known(file_status s) {
 639   return s.type() != file_type::status_error;
 640 }
 641
 642 bool is_directory(file_status status) {
 643   return status.type() == file_type::directory_file;
 644 }
 645
 646 error_code is_directory(const Twine &path, bool &result) {
 647   file_status st;
 648   if (error_code ec = status(path, st))
 649     return ec;
 650   result = is_directory(st);
 651   return success;
 652 }
 653
 654 bool is_regular_file(file_status status) {
 655   return status.type() == file_type::regular_file;
 656 }
 657
 658 error_code is_regular_file(const Twine &path, bool &result) {
 659   file_status st;
 660   if (error_code ec = status(path, st))
 661     return ec;
 662   result = is_regular_file(st);
 663   return success;
 664 }
 665
 666 bool is_symlink(file_status status) {
 667   return status.type() == file_type::symlink_file;
 668 }
 669
 670 error_code is_symlink(const Twine &path, bool &result) {
 671   file_status st;
 672   if (error_code ec = status(path, st))
 673     return ec;
 674   result = is_symlink(st);
 675   return success;
 676 }
 677
 678 bool is_other(file_status status) {
 679   return exists(status) &&
 680          !is_regular_file(status) &&
 681          !is_directory(status) &&
 682          !is_symlink(status);
 683 }
 684
 685 void directory_entry::replace_filename(const Twine &filename, file_status st,
 686                                        file_status symlink_st) {
 687   SmallString<128> path(Path.begin(), Path.end());
 688   path::remove_filename(path);
 689   path::append(path, filename);
 690   Path = path.str();
 691   Status = st;
 692   SymlinkStatus = symlink_st;
 693 }
 694
 695 error_code has_magic(const Twine &path, const Twine &magic, bool &result) {
 696   SmallString<32>  MagicStorage;
 697   StringRef Magic = magic.toStringRef(MagicStorage);
 698   SmallString<32> Buffer;
 699
 700   if (error_code ec = get_magic(path, Magic.size(), Buffer)) {
 701     if (ec == errc::value_too_large) {
 702       // Magic.size() > file_size(Path).
 703       result = false;
 704       return success;
 705     }
 706     return ec;
 707   }
 708
 709   result = Magic == Buffer;
 710   return success;
 711 }
 712
 713 error_code identify_magic(const Twine &path, LLVMFileType &result) {
 714   SmallString<32> Magic;
 715   error_code ec = get_magic(path, Magic.capacity(), Magic);
 716   if (ec && ec != errc::value_too_large)
 717     return ec;
 718
 719   result = IdentifyFileType(Magic.data(), Magic.size());
 720   return success;
 721 }
 722
 723 namespace {
 724 error_code remove_all_r(StringRef path, file_type ft, uint32_t &count) {
 725   if (ft == file_type::directory_file) {
 726     // This code would be a lot better with exceptions ;/.
 727     error_code ec;
 728     for (directory_iterator i(path, ec), e; i != e; i.increment(ec)) {
 729       if (ec) return ec;
 730       file_status st;
 731       if (error_code ec = i->status(st)) return ec;
 732       if (error_code ec = remove_all_r(i->path(), st.type(), count)) return ec;
 733     }
 734     bool obviously_this_exists;
 735     if (error_code ec = remove(path, obviously_this_exists)) return ec;
 736     assert(obviously_this_exists);
 737     ++count; // Include the directory itself in the items removed.
 738   } else {
 739     bool obviously_this_exists;
 740     if (error_code ec = remove(path, obviously_this_exists)) return ec;
 741     assert(obviously_this_exists);
 742     ++count;
 743   }
 744
 745   return success;
 746 }
 747 } // end unnamed namespace
 748
 749 error_code remove_all(const Twine &path, uint32_t &num_removed) {
 750   SmallString<128> path_storage;
 751   StringRef p = path.toStringRef(path_storage);
 752
 753   file_status fs;
 754   if (error_code ec = status(path, fs))
 755     return ec;
 756   num_removed = 0;
 757   return remove_all_r(p, fs.type(), num_removed);
 758 }
 759
 760 error_code directory_entry::status(file_status &result) const {
 761   return fs::status(Path, result);
 762 }
 763
 764 } // end namespace fs
 765 } // end namespace sys
 766 } // end namespace llvm
 767
 768 // Include the truly platform-specific parts.
 769 #if defined(LLVM_ON_UNIX)
 770 #include "Unix/PathV2.inc"
 771 #endif
 772 #if defined(LLVM_ON_WIN32)
 773 #include "Windows/PathV2.inc"
 774 #endif